This is the code in bodhi that looks up resultsdb for test results:
https://github.com/fedora-infra/bodhi/blob/develop/bodhi/server/templates/update.html#L17-L349

When looking at the queries it makes:
https://github.com/fedora-infra/bodhi/blob/develop/bodhi/server/templates/update.html#L343-L348

I think these are too specific to return Atomic CI test results.
We would like need more something along the lines of:
https://taskotron.fedoraproject.org/resultsdb/results?original_spec_nvr=<build NVR>&testcases=org.centos.prod.ci.pipeline.complete

Such as: https://taskotron.fedoraproject.org/resultsdb/results?original_spec_nvr=criu-3.3-5.f26&testcases=org.centos.prod.ci.pipeline.complete

One thing I found out while looking into this is that we seem to have not act on some messages:

The fedmsg consumer that loads messages into resultsdb in RDU2 has the following entries in its logs:

Sep 26 09:31:23 ci-cc-rdu01.fedoraproject.org fedmsg-hub[29757]: [2017-09-26 09:31:23][resultsdb_listener.pipeline_complete    INFO] PipelineCompleteConsumer: Done
Sep 26 09:31:23 ci-cc-rdu01.fedoraproject.org fedmsg-hub[29757]: [2017-09-26 09:31:22][resultsdb_listener.pipeline_complete    INFO] PipelineCompleteConsumer: Got a message u'org.centos.prod.ci.pipeline.complete': u'2017-6c9b1886-389b-40cb-b5d9-bc4538f2353d'
Sep 24 14:21:15 ci-cc-rdu01.fedoraproject.org fedmsg-hub[29757]: [2017-09-24 14:21:15][resultsdb_listener.pipeline_complete    INFO] PipelineCompleteConsumer: Done
Sep 24 14:21:15 ci-cc-rdu01.fedoraproject.org fedmsg-hub[29757]: [2017-09-24 14:21:15][resultsdb_listener.pipeline_complete    INFO] PipelineCompleteConsumer: Got a message u'org.centos.prod.ci.pipeline.complete': u'2017-1e274bdc-6591-4cc8-98b3-9a330c3ef55c'

I did not remove any thing, which means there was no org.centos.prod.ci.pipeline.complete loaded in the DB between September 24th at 14:21 UTC and September 26th at 09:31 UTC, but looking at datagrepper: https://apps.fedoraproject.org/datagrepper/raw?topic=org.centos.stage.ci.pipeline.complete there were messages on the 25th...

Actually, looking at these messages in datagrepper, they were all "topic": "org.centos.stage.ci.pipeline.complete",

I start to wonder if there hasn't been something not working on the pipeline itself :(

Ah, I see my datagrepper query was wrong, it was querying for the stage message.

This is the proper query: https://apps.fedoraproject.org/datagrepper/raw?topic=org.centos.prod.ci.pipeline.complete

There were still some messages on September 25th, though a bunch of them were failures and did not include the original_spec_nvr so we would not be able to map them to bodhi updates :(
(That includes the cockpit message: https://apps.fedoraproject.org/datagrepper/id?id=2017-a926d1e4-f6f5-4824-9dcb-a7d9b57db9a4&is_raw=true&size=extra-large )

not include the original_spec_nvr so we would not be able to map them to bodhi updates :(

This has been reported upstream: https://github.com/CentOS-PaaS-SIG/ci-pipeline/issues/394

I got an idea for this, since the messages have been caught by datagrepper we can teach https://pagure.io/ci-resultsdb-listener/ to retrieve it from there and add it to resultsdb. This is going to be an admin-only manual process but may help us to recover from time to time.

Integration in bodhi up for review: https://github.com/fedora-infra/bodhi/pull/1847

Screenshots are always nice, aren't they? :)

Looks like it is merged in bodhi

Has it been deployed?

Do we need to keep this issue open?